Establishing call-based audio sockets within a componentized voice server

• US 7,424,432 B2
• Filed: 04/28/2004
• Issued: 09/09/2008
• Est. Priority Date: 04/28/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Creamer, Thomas E., Moore, Victor S., Nusbickel, Wendi L., Sliwa, James J., Dos Santos, Ricardo
• Primary Examiner(s): McFadden; Susan
• Application Number:
  US10/833,516
• Publication Number:
  US 20050246175A1
• Time in Patent Office: 1,595 Days
• Field of Search: 704/270.1, 379/88.14
• US Class Current: 704/270.1
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

A method of interfacing a telephone application server and a speech engine can include the step of establishing one or more audio sockets in a media converting component of the telephone application server. The audio socket can remain available for approximately a duration of a call. A work unit that requires processing by a speech engine can be detected for the call. An identifier for the audio socket and a data for the work unit can be conveyed to a selected speech engine. Work unit results from the selected speech engine can be received by the media converting component via the previously established audio socket.

Claims

1. A method for establishing call-based audio sockets within a componentized voice server for communicating with external speech engines, the method comprising the steps of:

receiving a call;

initializing the voice server to handle the call;

establishing at least one audio socket in a media converting component of the voice server for interfacing the voice server with an external speech engine, said audio socket remaining available for approximately a duration of the call;

detecting a work unit of said call that requires processing by an external speech engine;

conveying an identifier for the audio socket and data relating to the work unit to a selected external speech engine; and

receiving work unit results from the selected external speech engine via the audio socket.

2. The method of claim 1, wherein the work unit is a speech utterance of a party participating in the call, and wherein the external speech engine is an automatic speech recognition engine.

3. The method of claim 1, wherein the work unit is a text segment generated by the voice server, and wherein the external speech engine is a text-to-speech engine.

4. The method of claim 1, further comprising the steps of:

after the work unit data is processed, the external speech engine initializing a communication connection with the audio socket; and

responsive to said initialization, conveying pending audio signals from the media converting component to the audio socket.

5. The method of claim 1, further comprising the steps of:

detecting a different work unit of said call that requires processing by an external speech engine;

conveying said identifier for the audio socket and conveying data relating to the different work unit to a different external speech engine; and

receiving work unit results from said different external speech engine via the same audio socket.

6. The method of claim 5, wherein said external speech engine is an automatic speech recognition engine and wherein said different external speech engine is a text-to-speech engine.

7. The method of claim 1, wherein said at least one audio socket is a plurality of audio sockets.

8. The method of claim 7, wherein said plurality of audio sockets comprises at least one input audio socket and at least one output audio socket.

9. The method of claim 1, wherein said componentized voice server has a componentized architecture of different modular components.

10. The method of claim 9, wherein each component within said architecture is configured to handle tasks in a functionally isolated fashion from other components of said architecture.